You come across the term Artificial Intelligence very often in the media, product descriptions and films. But what does it actually mean? Alan Turing, the creator of the term said about this. Artificial Intelligence is the science and engineering of making intelligent machines, especially intelligent computer programs. That doesn’t really help, this definition is too broad.
Scientists define intelligence as a way of planning, learning and reasoning that allows the use of natural languages. This is a very narrow definition and should be expanded to include the use of picture and sound.
Many software companies rename their software that can do seemingly intelligent things through data mining, etc. to Artificial Intelligence. That is commercially smart, of course, but not good for the understanding and certainly not for the image of what Artificial Intelligence is. That definition is therefore simply wrong.
The holy grail of Artificial Intelligence is General Artificial Intelligence. This means a self-reasoning and cognitive form. An intelligent computer where you can present a problem just like you would to a person. The AI itself searches for the data and makes connections when looking for solutions to a problem.
General AI is also usually the form you encounter in movies, but we are not there yet. For that reason we sometimes say, Artificial Intelligence does not exist. There are a number of distinguished appearances of AI that are already available.
- Computer Vision. The ability to have computers analyze video images and thereby recognize objects, people, relationships and even behavior.
- Machine Learning (ML). Computers can analyze data and independently discover connections. These therefore do not need to be pre-programmed. For example, you can predict housing prices
- Natural Language Processing (NLP). The ability of computers to “understand” natural languages such as Dutch and English. Understanding should not be taken literally, but based on sentence structure and words, computers can extract the essence of text and take action or respond accordingly.
- Sound recognition. This also includes voice recognition, but is not limited to this.
- Cognitive Analytics. We try to get computers to think, by linking words. In fact, this is a combination of NLP and ML. An example of this is that we once created a chatbot that could answer questions about topics on Wikipedia.
The combination of a number of the above technologies makes Artificial Intelligence special, mind-boggling and sometimes even a bit scary.
There are still a number of technologies that claim the term Artificial Intelligence, such as Robotics. We see this as an application that uses algorithms from the above spectrum.
So they do not necessarily add anything to making software more intelligent. Planning and expert systems are currently mainly a combination of technology. After all, is planning a separate cognitive process and therefore with its own algorithms? Or is it a matter of processing data by means of a number of well-known features. The moment a system can independently make a plan without programming, it falls into the AI domain.
The attached video is an example of an application made by Centillien that measures distance between people.
Recognizing Theft
Another nice example of Computer vision is the detection of theft. For a Dutch television program we trained our software Intra to recognize theft. That went like this
You can safely say that Artificial Intelligence is a complex subject. Hopefully the above explanation has given a little more clarity. Consider checking out the possible applications of computer vision for various sectors.
